Nvidia is in negotiations to provide an estimated financial guarantee of US$ 250 billion (equivalent to R$ 1.3 trillion) to OpenAI. This contribution is part of a massive data center construction project in the United States, representing one of the boldest financial moves in the current artificial intelligence (AI) competition in the country.
Data Center Complex Details
Sources informed to The Wall Street Journal indicated that this guarantee from the semiconductor manufacturer would enable OpenAI to lease a 10-gigawatt (GW) complex being developed in southern Ohio, USA, by SoftBank's energy division. Considering the chips needed to equip these data centers, the total cost of the undertaking could exceed US$ 500 billion (R$ 2.5 trillion), establishing it as the largest announced data center project.
Financing and Governance
The energy allocated to this project is supervised by the U.S. government and receives additional funding from Japan, according to a recent trade agreement between the two nations. U.S. Commerce Secretary Howard Lutnick is involved in determining which corporations will have access to the energy supply. Since OpenAI is a private entity that does not yet generate profit and lacks investment-grade credit rating, Nvidia's financial support would help the ChatGPT developer secure resources under more favorable conditions. The company has been conducting advanced negotiations for the complex lease for several weeks.
Interest from Other Companies
Besides OpenAI, other companies have expressed interest in the installation. According to a source, Anthropic, Microsoft, and Google have also spoken with Lutnick in recent weeks about the project.
Dynamics of the AI Race
The competition among AI companies for chips and sufficient electrical power to run their models has increased the appeal of large projects backed by governmental support or assured electricity supply. In the currently debated format, Nvidia would act as a guarantor for various financial instruments, aiming to give creditors greater assurance that the project's financing will be completed. However, the final terms have not yet been defined, and the agreement may not materialize.
Additional Financial Structure
The US$ 250 billion guarantee would cover the data center lease contract and the debt required for its construction, but it would not include the Nvidia chips that will be installed in the infrastructure. Concurrently, the semiconductor manufacturer is also negotiating a deal to finance the acquisition of these chips by OpenAI, a pact that could reach US$ 350 billion (R$ 1.8 trillion), according to people close to the talks.
Structural and Future Implications
This type of financial architecture, where different operations support each other, has raised concerns that the sector could become vulnerable if investor enthusiasm wanes or the growth of AI companies slows down. The campus will consume about 10 GW of electricity, enough to supply millions of residences. Construction will require several years, with the first phase expected to be completed in 2028, providing approximately 800 MW of electrical capacity.
Japan-US Partnership and Federal Land
The project also represents a significant investment for Lutnick and the administration of President Donald Trump. In fulfillment of Japan's commitment to increase investments in the United States in exchange for tariff reductions, Japan agreed to contribute US$ 33 billion (R$ 167.8 billion) to a natural gas power plant on federal land in Ohio. This facility will be managed by SoftBank Energy, a company controlled by SoftBank founder Masayoshi Son. Lutnick, Son, and U.S. Secretary of Energy Chris Wright participated in the groundbreaking ceremony in March, at which time SoftBank Energy received investments from OpenAI, and SoftBank itself is a major investor in the AI company.
Location Advantages and Revenue Model
By using federal land in Ohio, the state government and its officials seek to avoid hurdles related to environmental licenses and resistance from local communities, factors that have delayed similar projects in other parts of the country. The complex will be built on a former decommissioned uranium enrichment facility, located about 80 kilometers south of Columbus, USA. Despite this, the project depends on the completion of financing and the acquisition of all equipment to proceed fully.
Business Model and Projections
A source revealed that the United States government will pay a fee to SoftBank Energy for operating the plant. The revenue generated from energy sales will be split between Japan and the United States until the Japanese investment of US$ 33 billion is recovered; after that, the state government will receive 90% of the revenue. This arrangement illustrates the state administration's tactic of accelerating investments in crucial sectors through direct collaborations with the private sector.
Innovation in Infrastructure Financing
The proposed structure reflects an evolution in how large AI infrastructure projects are financed. Technology companies with high credit ratings have begun using their own balance sheets to facilitate access to credit for smaller companies, a practice known as 'credit wrapper'. Google, for example, has already guaranteed funds for some data centers used by Anthropic, partly to boost the sale of its own AI chips, the Tensor Processing Units (TPUs).
Impact on OpenAI and Market
If realized, the Ohio complex will be the first data center directly used by OpenAI as a lessee, bringing the company closer to controlling infrastructure that is currently mostly leased from cloud providers like Microsoft, Amazon, and Oracle. Recently, OpenAI increased its estimate of spending on computing capacity to about US$ 750 billion (R$ 3.8 trillion) by 2030, surpassing the initial projection of approximately US$ 600 billion (R$ 3 trillion) released earlier this year. It is not yet clear how the new agreement will influence these forecasts.
Market Outlook
Simultaneously, OpenAI and Anthropic are moving towards future Initial Public Offerings (IPOs), with extremely high market valuations, which intensifies the pressure for accelerated growth. Meanwhile, Nvidia, valued at about US$ 5 trillion (R$ 25.4 trillion), reported in its latest annual report that data center financing deals could decrease its short-term cash flow and increase its exposure to customer credit risk.
